Development of a model for estimation of wind farm production losses due to icing

Abstract: Wind turbines operating in cold climate are exposed to periods of icing which lowers the plantprofitability by affecting the annual production. The loss of production has two components:The first (and most important) component is reduced power during operation due to disturbedaerodynamic properties of the blades. The second component is increased standstill. During this thesis project, methods to estimate production losses of a wind farm due to icinghave been developed, as well as a software tool to facilitate the use of these methods and thepresentation of the results. A method based on common metrological data and availableproduction data was desired, as modelling ice-related losses is expensive and may be inaccurate. The methods developed are based on using measured data for each turbine, such as activepower, temperature, wind direction and wind speed, and through this data describe theindividual turbine’s performance during different conditions. Production losses were thenestimated by comparing actual and expected power output (for the given wind speed). Thethesis then expanded on this basic concept by using reanalysis and mesoscale modelled data,which offers greater variety in the way estimating the losses may be performed, as well as theoption to derive losses for periods not covered by the production data. It was also important to develop a flexible and portable method that could incorporate newseasons of data or estimate losses for different wind farms with a completely differentconfiguration of turbines. The methods are developed using data from a wind farm in northern Sweden, consisting of 40Vestas V90 turbines and constructed a few years ago. It was found that eastern position in the wind farm and turbine altitude correlates with higherice-related losses, and that easterly winds relate to higher such losses than westerly winds. Thelosses during operation were estimated to 6.4 % of annual possible production and stops due toicing to 2.1 % of the total time. The losses figures are comparable to an earlier study performedin 2011 based on the same wind farm. The possibility of anti- or deicing systems for the wind farm and the profitability of such aninvestment should be further investigated as the wind farm is expected to continue operation fortwenty years or more.